We have always had a checklist for reviewing financial statements, these days datasnipper helps with some of those existing tasks.

In terms of checking the results, we have a policy to tell teams they must check the inputs of each cast/internal consistency suggestion to make sure the sums are complete for example. This is still much quicker than trying to add up the numbers manually but minimises reliance on the system to pick things up correctly.

God I hate a checklist and they're the easiest way to make people switch off their brains! You just need to train people to critically appraise the results. Alternatively train an AI agent to do the review...

I agree that we may have too many checklists, but in this instance it is valid and necessary. We use the checklist to ensure that the other components of the financials are complete, such as Pentana and disclosures. I think the is high risk to give this to AI to do at this stage as this is ultimately the final product our client sees and where we need to apply our expertise.

I’m not sure there’s any expertise required to complete a checklist - it’s why most firms push it down to junior staff… The expertise lies in knowing the accounting standards etc and if you have that you don’t need a checklist to undertake a financial statement review. I’ve seen a couple of firms use AI to complete checklists and it’s way more accurate than a human

Agreed that the checklist itself is not difficult but confirming for example the correct disclosures are in place, especially for larger listed companies, is more complex. The checklist, for me, is there to ensure we cross the t's and dot the i's. FSS does of course go some way to assist in the review, but to ensure that there isn't omissions is difficult, to ensure supplementary information is not contradictory is difficult to pass on to automation. The application of the correct accounting framework etc. Or perhaps I need to see more evidence in practice to be convinced. 

Thanks all :)